Pyrolysis of Potassium-Doped Wood at the Centimeter and

Oct 15, 2015 - Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York 14853, United States. § School of Energy a...
1 downloads 11 Views 2MB Size
Subscriber access provided by Stockholm University Library

Article

Pyrolysis of potassium-doped wood at the centimeter and sub-millimeter scales Mishal Harshad Shah, Lei Deng, Hayat Bennadji, and Elizabeth M. Fisher Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.5b01776 • Publication Date (Web): 15 Oct 2015 Downloaded from http://pubs.acs.org on October 18, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34

Energy & Fuels

1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Pyrolysis of potassium-doped wood at the centimeter and sub-millimeter scales Mishal H. Shah§,%, Lei Deng+,%, Hayat Bennadji%,1, Elizabeth M. Fisher%* §

Department of Chemical Engineering Loughborough University,

Loughborough, Leicestershire, LE11 3TU, UK %

Sibley School of Mechanical and Aerospace Engineering Cornell University Ithaca, NY 14853, USA +

School of Energy and Power Engineering Xi'an Jiaotong University Xi'an, 710049, China

ABSTRACT The effect of potassium additives on pyrolysis of poplar was investigated at 427 ○C, both at the sub-millimeter scale, through thermogravimetric analysis (TGA), and at the centimeter scale, through pyrolysis of wood cylinders in a turbulent reactor. Internal temperatures and timeresolved rates of production of gases and light volatiles were measured in the centimeter-scale study. The potassium level in the samples was varied through vacuum treatment with distilled water or solutions of KCl or K2CO3, resulting in potassium levels of approximately 100 ppm, 4500 ppm, and 7000 ppm (dry, mass basis). At the centimeter scale, potassium addition had a dramatic effect on conversion time and on the magnitude of exothermic temperature excursions, and a significant effect on the yields of gases and light volatiles. Consistent with literature, sub1

Current Address: Energy Coast & Environment, Louisiana State University, Baton Rouge, LA 70808, USA

ACS Paragon Plus Environment

Energy & Fuels

Page 2 of 34

2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

millimeter-scale TGA experiments with external temperature control also indicated a catalytic effect of the potassium additives, with K2CO3 more effective than KCl in promoting pyrolysis and char formation. INTRODUCTION Experiments at the sub-millimeter scale have shown that mineral content of biomass has a strong effect on the pyrolysis process, and thus on the thermochemical conversion of biomass to heat, useful energy carriers, and/or other useful products. Thermal conversion processes start with the pyrolysis or devolatilization of biomass, producing a range of gaseous, condensable, and solid products, which are known generically as gas, tar, and char. The proportions, conversion rates, and detailed composition of these product streams depend on many operational variables and biomass characteristics including mineral content. Alkali metals, especially potassium, are among the most abundant of inorganics present in biomass.1 Potassium levels in biomass vary over a wide range depending on biomass type and growing conditions;1 for example Baxter et al.2 report potassium mass fractions of 0.4% in wood-derived fuels, 2% in straws, and 3% in almond hulls; 80-90% of this potassium is in water soluble or ion exchangeable forms. The literature indicates that higher potassium levels in biomass are generally associated with lower temperatures required for pyrolysis, lower levels of tars, and higher levels of gases and chars in the product stream,3-7 While slow pyrolysis experiments follow these trends, some fast pyrolysis experiments show different behavior 8-10. The impact of potassium on the yields of specific products and classes of products has also been documented.3,10-15 Catalytic effects of potassium have been demonstrated at levels as low as 0.05% potassium by mass. 15

ACS Paragon Plus Environment

Page 3 of 34

Energy & Fuels

3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Numerous studies have compared low- and high-potassium versions of the same powdered biomass material, in order to isolate the potassium effects. In some of these studies, differences in potassium loading are achieved through washing away naturally occurring potassium, using acid or hot water.3,,8,16-18 In others, higher levels of potassium are achieved by mechanically mixing potassium salts with powdered biomass,19 by soaking biomass powder in potassium salt solutions,16 or by multistep ion exchange processes.20 With a few exceptions,8 the effect of potassium addition or removal on pyrolysis is investigated using sub-millimeter-scale biomass particles, either in fluidized beds9 or in thermogravimetric measurements where the temperature of sub-millimeter-scale biomass particles is externally controlled during slow heating 3,21. While these experiments are useful, they do not provide information about the influence of thermochemistry and intraparticle secondary charring reactions. The current study takes a different approach to doping biomass that makes it possible to adjust potassium content in centimeter-scale biomass samples. Thus we are able to pyrolyze low- and high-potassium biomass samples at the centimeter scale, altering the pyrolysis kinetics while having a minimal impact on heat transfer characteristics. The centimeter and larger scale is important in devices such as fixed bed gasifiers, and has proven useful for elucidating the effects of thermochemistry.21 Although the impact of potassium on kinetics should remain the same as in sub-millimeter experiments, the importance of the kinetics may be substantially reduced in the case of heat-transfer-limited pyrolysis in larger particles. The relative speeds of chemical kinetics, internal heat transfer, and external heat transfer are typically described in terms of the Pyrolysis and Biot numbers, Py and Bi.22,23 Under the conditions tested, the Biot number is approximately 5, indicating a thermally thick condition with significant internal temperature gradients. The Pyrolysis number, based on cellulose

ACS Paragon Plus Environment

Energy & Fuels

Page 4 of 34

4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

kinetics22 at the final temperature, is below 0.01, indicating that the conversion time should be determined by the heat transfer process, which is much slower than the chemical kinetics. Thus, on the basis of the value of Py, we would anticipate the catalytic effect of the potassium on kinetics to have minimal impact on conversion times. The results of the current study, which contradict these expectations, indicate some problems with using the Pyrolysis number to characterize the behavior of large particles. MATERIALS, APPARATUS, AND METHODS Chemicals and preparation of wood samples Potassium chloride (purity ≥99.5%) and potassium carbonate (purity ≥99%), were purchased from Sigma Aldrich, St. Louis, MO and were used to prepare solutions with distilled water; mass fractions are given in Table 1. A single dowel of dried, untreated poplar, free from visible knots and heartwood, was purchased from Lowes Home Improvement, Ithaca, NY. The dowel was cut into cylinders of the specified length and dried to completion at 105 ○C in air. Next, in sets of seven, the cylinders were treated with either distilled water or salt solutions, as follows. The cylinders were immersed in 2.5 liters of the water or solution and held below the liquid surface with weights. The pressure in the container was lowered to 2450 Pa with a vacuum pump, and that pressure was maintained for 5 hours, i.e. until bubbles had essentially stopped emerging from the wood. The treated cylinders were then dried in air in a process developed by trial and error as a tradeoff between cracking/splitting (at high drying rates) and mold growth (at low drying rates). For the first 90-140 hours, the cylinders were stored on end in a small beaker inside a dessicator. Approximately once a day, the cylinder orientation was reversed and fresh Drierite® dessicant was placed in the dessicator. During this stage, the cylinders lost

ACS Paragon Plus Environment

Page 5 of 34

Energy & Fuels

5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

approximately 8.4% of the added water mass per day. Upon first appearance of tiny dots of mold, the cylinders were transferred to an oven and dried at 40 ○C for 24 hours, then dried at 105 ○

C to completion. All drying took place in air. No visible cracking or splitting occurred with

this drying regimen. Treated cylinders were stored in a dessicator until used. Analysis of mineral content Samples were prepared for mineral analysis as follows: a cylinder was held in a lathe, and specific sections of it were cut into shavings with a cutting tool or drill bit. All surfaces in contact with the wood had been cleaned with alcohol and dried before use, and the shavings were collected on fresh aluminum foil. Dry shavings (mass 0.14 to 1.5 gram) were weighed and then digested with nitric and perchloric acid in an automated Vulcan 84 digestion unit (Questron Technologies, Mississauga Canada). Digestion took place over the course of 3 hours, with a controlled temperature program with a maximum temperature of 180 ○C. Yttrium was added to the mixture before digestion, for use as an internal standard. The solutions were analyzed for their elemental composition by inductively coupled plasma atomic emissions spectroscopy (ICP-AES) using an axial viewed Spectro Arcos ICP-AES (Kleve, Germany). ICP measurements of potassium and other minerals were performed for two different cylinders treated with each doping treatment, and reported on a dry basis. For the first cylinder, spatially resolved ICP readings were obtained by analyzing shavings from different regions of the cylinder. Averages for the entire cylinder were then obtained by averaging the spatially resolved results, weighted by the volume of the sampled region. For the second cylinder, the ICP

ACS Paragon Plus Environment

Energy & Fuels

Page 6 of 34

6 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

measurements were performed on a powder obtained by converting the entire cylinder to shavings, which were then ball-milled and sieved. Sub-millimeter-scale pyrolysis experiments Wood shavings from an entire cylinder were converted to powder using a ball mill, then sieved to retain particles below 250 microns. Experiments were performed in duplicate with sample masses between 6.4 and 14.2 mg, using platinum crucibles in a TA Instruments (New Castle, DE) Q500 thermogravimetric analyzer (TGA). The temperature profile was: 20 ○C/min to 105 ○

C; hold at 105 ○C for 10 minutes; 5 ○C /min to 420 ○C; hold at 416 ○C for 10 minutes. Sample

and balance purge flows were N2 with flowrates of 60 and 40 standard cubic centimeters per minute, respectively. Masses were referenced to the mass at the end of the 105 ○C plateau, i.e. presented on a dry basis. The rate of change of mass was determined without smoothing, using first-order forward differencing. Centimeter-scale pyrolysis experiments Each wood cylinder was pyrolyzed individually by insertion into a turbulent nitrogen flow that had been preheated to 427 ○C. Details of the pyrolysis method and measurement techniques are described elsewhere;24 key features are outlined here. During pyrolysis, temperatures at two locations inside the wood were monitored with 0.5-mm-OD sheathed type K thermocouples. Products were sampled continuously through the gas cell of an online FTIR system, which was used to quantify gaseous and light volatile products. An NDIR system was also used to quantify carbon monoxide with better time resolution. After pyrolysis, the cylinder was withdrawn into an unheated side arm of the reactor and cooled with a cold nitrogen flow until its temperature dropped below 100 ○C, when it was removed and placed in a dessicator. The choice of reactor

ACS Paragon Plus Environment

Page 7 of 34

Energy & Fuels

7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

temperature and the unusual reactor design produce initial heat fluxes on the order of 20 W/m2, with rapid dilution and no extra-particle reaction of product gases. The apparatus, procedure, and operating conditions, including nitrogen and sampling flowrates, were as described previously24 with the following exceptions: 1) the cylinder was held in the high-temperature flow for only 20 minutes; 2) turbulizers had been added to the reactor upstream of the wood insertion point, and 3) thermocouple positions in the wood were different, and a third thermocouple monitored the gas temperature near the wood, providing an accurate indication of the time of insertion. RESULTS AND DISCUSSION Wood and cylinder characteristics At the end of the entire treatment process, the cylinders’ diameters were 1.889 ± 0.009 cm, and their lengths were 4.010 ± 0.001 cm, essentially unchanged from the start of the treatment. The density of the wood after washing with distilled water and drying was 523.8 kg/m3, and the mass fractions C, H, O, and N of the water-treated wood were 54.9%, 6.21%, and 46.4%, and 0.09%, respectively, on a dry basis, as determined with an NC2500 elemental analyzer. The porosity of the wood, ߝ, was estimated from mdry and mwet, the mass measured immediately before and after soaking in water or salt solution, respectively: ߝ=

݉௪௘௧ − ݉ௗ௥௬ ߩ௟௜௤௨௜ௗ ܸ௖௬௟௜௡ௗ௘௥

This value obtained, 71%, agrees reasonably well with the value obtained from the measured dry density and the theoretical density of void-free wood,25 which was 65%. Potassium loading: average and spatial distribution

ACS Paragon Plus Environment

Energy & Fuels

Page 8 of 34

8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Average potassium levels in the doped wood were determined in several different ways, giving generally consistent results, as seen in Table 1. First, potassium levels were estimated by comparing the dry mass before doping to the wet mass of the cylinder immediately after doping, and assuming that the mass of liquid added to the cylinder had the same composition as the initial aqueous solution. Next, masses of the dried treated cylinders were compared to the dry masses before doping, and K fraction was calculated on the assumption that the mass added to the cylinder was either KCl or K2O. Note that treating the cylinders with distilled water resulted in minimal net change in mass (loss of 0.04% of the initial mass). Finally the mineral mass fractions of K and other minerals were analyzed by ICP for selected cylinders, as described above. In all cases, the average concentration of potassium was similar in the K2CO3 and high-KCl samples, and was approximately 40% lower in the low-KCl samples, consistent with the potassium concentrations of the solutions used in preparing the samples. The potassium levels achieved were comparable to those in high-mineral biomass. Note that the change in dry cylinder mass gives consistent results only when it is assumed that the K2CO3-doped samples lost the added CO2 during the drying process. Although dilute K2CO3 solutions absorb CO2 from the gas phase,26 the process can shift to desorption as evaporation proceeds and produces more concentrated solutions. Potassium content of the untreated poplar wood was not measured, but the levels of potassium in the distilled-water cases are reduced approximately tenfold from typical literature values for hardwoods.27 However, because of mass-transfer limitations in the wood cylinders, the distilled water treatment did not remove minerals to the very low levels typically achieved in washing of

ACS Paragon Plus Environment

Page 9 of 34

Energy & Fuels

9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

powdered biomass (typically 2-40 ppm).13

Note that the mass fractions of other minerals are

listed in the Suppporting Information. A representative spatial distribution of potassium (measured for a K2CO3-doped cylinder) is shown in Figure 1, with the width of the symbol representing the axial extent of the spatial region (annulus or cylinder of equal volume) from which the shavings were made. Note that spatially resolved data is not available in one axial region, which was used for holding the sample in the lathe. The spatial distribution of the potassium is highly nonuniform, with much higher levels near the surfaces where the drying occurred, namely at the axial positions closest to the upper end of the cylinder, and at the radially outer positions. Similar results have been observed by Smith and Cockcroft,28 and appear to be due to the transport of salt-laden liquid during the drying process. In the future, if a more uniform loading is desired, larger wood cylinders could be doped and dried, then machined down to remove the outer region where most of the spatial variability occurs. Spatial distributions for the other cylinders show similar patterns. Even in the distilled-water case, naturally-occurring potassium was redistributed during drying to create a similar spatial pattern. Sub-millimeter-scale pyrolysis experiments Figure 2 shows the results of TGA experiments. For each wood treatment, results from two runs are overlaid in the same color. The duplicate TGA results are virtually indistinguishable on the scale of the figure. This agreement, despite large differences in initial sample mass (e.g. 12 vs. 6.4 mg), indicates that the sample was under kinetic control during pyrolysis. If mass transfer limitations had been important, then results with different sample masses would have differed significantly.

ACS Paragon Plus Environment

Energy & Fuels

Page 10 of 34

10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2 shows a substantial impact of the potassium treatment on the pyrolysis kinetics, as observed previously. The water-treated wood shows typical behavior, with a maximum at 349 ○

C, and a shoulder around 275 ○C, consistent with the decomposition of cellulose and

hemicellulose7. The wood treated with high and low KCl concentrations shows qualitatively similar behavior, but with all features shifted to lower temperatures. The shift of the cellulose peak, to 309 and 312 ○C for the two treatments, is greater than the shift of the shoulder, to about 260 ○C.

The K2CO3-treated wood pyrolyzes at an even lower temperature, with its peak rate of

mass loss occurring at 297 ○C. For this treatment, it appears that the hemicellulose and cellulose peaks have merged; however, a small shoulder is visible at about 220 ○C. Char levels measured at the end of the pyrolysis process at 416 ○C are: 14.3%, 24.0%, 25.3%, and 27.2%, on a dry basis, respectively, for wood treated with distilled water, low KCl, high KCl, and K2CO3. The TGA results are consistent with the literature. The peak decomposition temperature is 10-15 ○

C lower in the water-treated samples of the current study than in the water-washed hardwood

samples of Mueller Hagedorn et al. 13 at the same heating rate. This difference may be due to the somewhat higher K levels in the water-treated samples than in Mueller-Hagedorn’s washed biomass7, or it may simply reflect differences in the starting biomass composition. Consistent with the present study, Mueller-Hagedorn13 observed substantial reductions in the temperature of peak decomposition with potassium doping of 5000 ppm. Also consistent with the current study are previous observations29 that catalytic effects are not sensitive to the differences in potassium loading level in the range 4000-54,000 ppm. Other researchers have specifically examined KCl and K2CO3 additives, and the differences in their effects on pyrolysis. At mass fractions similar to those in the present study, Liu et al.30 found that both KCl and K2CO3 decreased the temperature at which pyrolysis of cellulose

ACS Paragon Plus Environment

Page 11 of 34

Energy & Fuels

11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

occurred, and both promoted the formation of char. K2CO3 was somewhat more effective catalytically than the same K loading of KCl. Yang et al.31,32 investigated the effect of minerals on pyrolysis of biomass components (cellulose, hemicellulose, and lignin), mixtures of components, and biomass. Different mineral additives, including KCl and K2CO3, were mixed directly with the components at very high loadings (10% by mass), and thermogravimetric analysis was performed. Neither KCl nor K2CO3 influenced lignin pyrolysis behavior in the temperature range of the present study. For K2CO3, there was a promoting effect on cellulose pyrolysis and an inhibiting effect on hemicellulose pyrolysis, in both cases producing a DTG trace that seemed to show two decomposition processes occurring at different temperature ranges. KCl had little effect on the temperature of peak decomposition, but reduced the peak mass loss rate, thus producing more char. In contrast, Raveendran et al.4 reported only small impact of the anion on char yields for washed biomass samples impregnated with different potassium salts. Di Blasi et al.14 investigated the effects of added NaOH, NaCl, K2CO3, and other alkali salts on the pyrolysis of washed fir wood. They found, as in the current study, that the additives with the lower pH tended to have a greater catalytic effect on pyrolysis. Centimeter-scale pyrolysis experiments Figure 3 shows typical temperature histories at the center of the wood cylinder. Each different line corresponds to a different wood cylinder pyrolysis experiment, with green lines corresponding to doping with the high-KCl solution and black lines corresponding to water treatment. The run-to-run differences among green lines or among black lines are due to (1) small differences in the placement of the thermocouple tip, and (2) continuing small changes in temperature of heated gases and reactor walls even after the system’s four-hour warm-up time. In the case of the KCl-doped experiments, two of the runs also show abrupt changes in the slope

ACS Paragon Plus Environment

Energy & Fuels

Page 12 of 34

12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

of the temperature history (around 220 s after insertion). These changes are apparently due to splitting of the wood and the exposure of the thermocouple to direct heat transfer to the turbulent nitrogen flow. Upon removal from the reactor, runs with abrupt slope changes were found to have split char samples and/or missing parts of the char. When splitting occurred, it invariably exposed a thermocouple. Temperature plots in Figure 4 show average temperature histories for a given treatment and thermocouple location. In that figure, solid lines are measured at the center of the cylinder, while dotted lines are measured at L/4, R/2. The jagged appearance of the centerline temperature history for K2CO3-doped and high-KCl-doped wood appear to be due to wood splitting. As previously observed for undoped wood samples21,24,33,34, the temperature histories show complex behavior because of the interaction between (1) heating by the nitrogen flow and walls of the reactor, which tends to bring the sample to thermal equilibrium with the external gas, and (2) endo- and exothermic reactions within the pyrolyzing wood. Because external heating is less important at the center of the particle, the center temperature histories show the greatest extremes of slope. Similarly, the faster pyrolysis of the doped wood samples allows less time for heat transfer; thus these samples are further from thermal equilibrium with the outside gas temperature and show greater effects of reaction thermochemistry. Figure 4 shows that samples with all four wood treatments have very similar temperature histories at temperatures below about 200 ○C, i.e. before reactions become significant. This similar low-temperature behavior implies that the potassium additives have minimal effect on the heat transfer properties of the samples. Above 200 ○C, the temperature histories start diverging: first K2CO3-doped sample rising more rapidly than the other three, then, at a slightly higher temperature, the two KCl-doped samples separating from the water-treated sample. All four

ACS Paragon Plus Environment

Page 13 of 34

Energy & Fuels

13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

samples show inflection points and maxima, but the locations of the maxima are affected by cracking/splitting in the K2CO3-doped and high-KCl-doped samples. The maxima of the three potassium-doped cases occur more than two times sooner than the maximum of the water-treated sample; the kinetics is fastest for K2CO3-doping, then high-KCl, then low-KCl. The timing of the temperature maximum is similar for the thermocouple at R/2, L/4 and the center thermocouple. Following the analysis of Di Blasi et al.,21 it is interesting to examine the time derivative of the center temperature, or “heating rate”, in Figure 5. After reactions start around 100 seconds, the time derivative of temperature shows somewhat different behavior for the four different wood treatments. For water-treated and low-KCl-doped samples, there are two peaks separated by a minimum, the same pattern observed by Di Blasi and coworkers for undoped samples of beech wood.21 For the high-KCl-doped sample, the two peaks start to merge into a peak and a shoulder, and for the K2CO3-doped sample, there is a single peak, cut short by splitting of the sample. Although the timing of the peaks and minimum are different for the different samples, the temperatures at which these features occur are comparable for the different treatments, and also compatible with the observations of Di Blasi and coworkers for undoped wood.21 The two peaks, occurring at approximately 320-330 ○C and 400-425 ○C, are attributed to exothermic decomposition of hemicellulose and lignin, respectively. 21 The minimum in heating rate is attributed to the combined effect of endothermic decomposition of lignin and cellulose. 21 The merging of these processes in the heating rate profiles reflects the merging of the mass loss peaks in the TGA profiles (Figure 2), as the catalytic effects of the potassium addition become stronger. Note that the initial heat flux estimated for the current experiments, 22 kW/m2, is comparable to that experienced in the Di Blasi et al. 21 experiments.

ACS Paragon Plus Environment

Energy & Fuels

Page 14 of 34

14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Potassium addition has a strong effect on the generation of gases and volatiles. Figures 6 through 9 show the time histories of the rates of generation of the species measured by FTIR. Each plot shows data from multiple samples, generally with very good consistency from sample to sample. The potassium additives affect both the timing of gas generation (much shorter for Kdoped samples) and the composition of the gas mixture. A consistent progression from watertreated, to KCl-treated, and finally to K2CO3-treated samples, is observed for most species. The two different loadings of KCl produce essentially indistinguishable species histories, as can be seen by plotting them together (not shown here). Table 2 lists integrated yields on a dry, additive-free basis. Potassium addition is associated with higher char and gas yields. For a given wood treatment, char yields are also higher than seen in the corresponding TGA cases, probably reflecting the occurrence of secondary charring reactions in the centimeter-scale samples. For most species, yields for the two KCl-doped samples are very similar to each other. Generally, the water-doped samples have one extreme of yields and the K2CO3-doped samples have the opposite extreme. Yields of CO2, CO, and methanol increase as potassium is added, while yields of formic acid and formaldehyde decrease.

Changes in

yields with additive loading may be due to additives’ effects on temperature histories and secondary tar reactions35 as well as their effect on pyrolysis kinetics. Because methanol, CO, and CO2 yields increase with temperature in the current range of conditions36, their increased yields with K may actually be due to the higher maximum temperatures encountered in doped samples. Because formic acid yields decrease with temperature and also with particle size,36 its decreased yield with K addition may be due to secondary reactions or to higher peak temperature. It is impossible to separate out the effects of these three modes of action with the information available. However, the timing of the appearance of CH4 (coincident with the brief

ACS Paragon Plus Environment

Page 15 of 34

Energy & Fuels

15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

high-temperature excursion) supports attributing the variation of its yield to the differences in temperature history. Figure 10 shows several characteristic times for the pyrolysis process. The 50% and 95% point for evolution of measured gases were interpolated from the species histories for each wood treatment. Note that the time scale for tar production is believed to be similar to that for gas production, as both products are derived from the same feedstock materials, and thus the depletion of the feedstocks marks the end of the devolatilization process for both gases and tars. The other characteristic times were based on the positions of maxima of temperature at the two measurement locations. The key observation from Figure 10 is that changing the kinetics through potassium doping has a substantial effect on all the characteristic pyrolysis times. Using each measure, the pyrolysis times of the water-washed samples are the longest, followed in order of decreasing pyrolysis time by the low-KCl, high-KCl, and K2CO3 samples. The ratio of slowest to fastest pyrolysis time varies from 1.6 to 2.45, depending on which measure is chosen, and in most cases the water-treated sample’s time is much longer than the other three samples’ times. The differences in conversion time appear to contradict the very low values of both internal and external pyrolysis numbers (Py), both estimated to be below 0.01 on the basis of cellulose kinetics evaluated at the external temperature. Note that similarly low values of Py were estimated by Bruch and Peters33 for pyrolysis centimeter-scale wood particles at this temperature. On the basis of the low Py numbers, it was expected that changes in kinetics would have minimal effects on conversion times, because heat transfer was the limiting process. The samples studied here have essentially identical heat transfer characteristics but different kinetics.

ACS Paragon Plus Environment

Energy & Fuels

Page 16 of 34

16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

This discrepancy exposes the difficulties of describing the pyrolysis process, especially on the centimeter scale, with values of a few dimensionless numbers. Most importantly, the use of the external temperature for evaluating the kinetic rate is not reasonable in the current experiment. 50% gas evolution occurred when the centerline temperature was only 242-309 ○C and when the more relevant temperature at L/4, R/2 was at 357-371 ○C. These temperatures, substantially below the external temperature of 427 ○C, allow a greater role for kinetics in determining the devolatilization timescale. Thermochemistry appears to play an important role in ensuring that substantial fractions of the mass loss process occur at low temperatures: once endothermic pyrolysis processes start, they reduce the rate of temperature rise of the sample. Furthermore, the different kinetics of different important reactions may lead to very different values of Py depending on which reaction is selected.34 It is clear that in this regime, pyrolysis time is determined by the interplay between heat transfer and chemistry. Figure 10 also shows that the exothermic temperature excursion occurs near the end of the devolatilization process. This finding is in agreement with previous observations with for undoped solid wood particles21,24,33 but not with observations for undoped packed beds.21 Finally, the form of the potassium and the potassium mass fraction, in the range tested, have relatively minor effects on characteristic times; the main difference is between the water-treated samples and the potassium-treated ones, as seen in the TGA results. The insensitivity of temperature histories, species yields, and characteristic times to the level of potassium present in the range of the doped samples, suggests that the nonuniformity of the potassium distribution in our samples may not have affected results significantly. CONCLUSIONS

ACS Paragon Plus Environment

Page 17 of 34

Energy & Fuels

17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Here we present what we believe are the first measurements of the effect of potassium doping on centimeter-scale wood particles. These measurements are relevant to the behavior of engineering devices (e.g. updraft gasifiers or chip combustors), where biomass chips undergo thermal processes. Kinetically controlled sub-millimeter-scale TGA measurements, with externally imposed temperatures, are provided for comparison. In the TGA experiments, potassium addition accelerated mass loss at a given temperature and tended to merge the peaks attributed to hemicellulose and lignin. Char yields increased with potassium addition at both centimeter and sub-millimeter scales. The strong accelerating effect of potassium on the pyrolysis process was present at the centimeter-scale as well, even though pyrolysis occurred in the thermally thick regime.

The

pyrolysis of the centimeter-scale particles yielded information about the thermochemistry of the different stages of the pyrolysis process. For the milder cases of potassium catalysis, two exothermic processes separated by an endothermic process could be observed, in a pattern and in temperature ranges similar to those observed previously for undoped wood. Thus, although the kinetics were faster than for the water-treated sample, the thermochemistry appeared qualitatively similar. At higher dopant loading and for the more effective catalyst (K2CO3), the processes appeared to merge, but the observations were hindered by the cracking or splitting of the wood samples. Species histories for centimeter-scale particles showed large change in the timing of gas and volatile evolution, as well as distinct increase in yields for certain species and decreases for others with potassium addition. Spatially resolved modeling is needed to gain a better understanding of the interactions between thermochemistry, kinetics, and transport in the potassium-doped system.

ACS Paragon Plus Environment

Energy & Fuels

Page 18 of 34

18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biomass materials naturally contain potassium and other catalytic materials at a variety of mass fractions. The present study indicates that the potassium level can have a strong effect on the pyrolysis time for these materials, even when they are pyrolyzed at the centimeter scale. SUPPORTING INFORMATION Mineral content of biomass samples ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of Yossie Hollander and Fondation des Fondateurs, and the State Scholarship Fund of China Scholarship Council. This work made use of the Cornell Center for Materials Research Facilities supported by the National Science Foundation under Award Number DMR-1120296. The authors thank Dr. Anthony Condo, Mr. Akio Enders, Mr. Dana Paul, Dr. Michael Rutzke, Prof. William B. Smith of SUNY ESF, and Ms. Colette Trouillot. REFERENCES (1) Vassilev, S. V.; Baxter, D.; Andersen, L. K.; Vassileva, C. G. Fuel 2010, 89, 913-933. (2) Baxter, L. L.; Miles, T. R.; Miles, T. R., Jr.; Jenkins, B. M.; Milne, T.; Dayton, D.; Bryers, R.W.; Oden, L. L. Fuel Proc. Technol. 1998, 54,47-78. (3) Nowakowski, D. J.; Jones, J. M.; Brydson, R. M. D.; Ross, A. B. Fuel 2007, 86, 2389-2402. (4) Raveendran, K.; Ganesh, A.; Khilar; Kartic C. Fuel 1995, 74, 1812-1822. (5) Fahmi, R.; Bridgwater, A.V.; Darvell, L.I.; Jones, J. M.; Yates, N.; Thain, S.; Donnison, I. S. Fuel 2007, 86, 1560-1569. (6) Fahmi, R.; Bridgwater, A.V.; Donnison, I.; Yates, N.; Jones, J. M. Fuel 2008, 87, 1230-1240.

ACS Paragon Plus Environment

Page 19 of 34

Energy & Fuels

19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(7) Oasmaa, A.; Solantausta, Y.; Arpiainen, V.; Kuoppala, E.; Sipala, K. Energy Fuels 2010, 24, 1380-1388. (8) Yoshida, T.; Turn, S.Q.; Yost, R. S.; Antal, M. J. Industrial & Engineering Chemistry Research 2008, 47, 9882-9888. (9) Morgan, T. J.; Turn, S.Q.; George, A. PLoS ONE 2015, 10, e0136511. Doi 10.1371/journal.pone.0136511; (10)

Mourant, D.; Wang, Z.; He, M.; Wang, X. S.; Garcia-Perez, M.; Ling, K.; Li, C. Z. Fuel

2011, 90, 2915-2922. (11)

Trendewicz, A.; Evans, R.; Dutta, A.; Sykes, R.; Carpenter, D.; Braun, R. Biomass

Bioenerg 2015,74,15-25. (12)

Patwardhan, P. R.; Satrio, J. A.; Brown, R. C.; Shanks, B. H. Bioresource Technol. 2010,

101, 4646-4655. (13)

Mueller-Hagedorn, M.; Bockhorn, H.; Krebs, L.; Mueller, U. J. Anal. Appl. Pyrolysis

2003, 68-69, 231-249. (14)

Di Blasi, C.; Galgano, A.; Branca, C. Ind. Eng. Chem. Res. 2009, 48, 3359-3369.

(15)

Eom, I. Y.; Kim, J. Y.; Kim, T. S.; Lee, S. M.; Choi, D.; Choi, I. G.; Choi, J. W.

Bioresour. Technol. 2012, 104, 687-694. (16)

Hwang, H.; Oh, S.; Cho, T. S.; Choi, I. G.; Choi, J. W. Bioresour. Technol. 2013,150,

359-366. (17)

Yang, H.; Yan, R.; Chen, H.; Zheng, C.; Lee, D. H.; Liang, D. T. Combust. Flame, 2006,

146, 605-611. (18)

Deng, L.; Zhang, T.; Che, D. Fuel Process. Technol., 2013,106, 712-720.

(19)

Wang, Z.; Wang, F.; Cao, J.; Wang. Fuel Process. Technol. 2010, 91, 942-950.

ACS Paragon Plus Environment

Energy & Fuels

Page 20 of 34

20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(20)

Richards, G. N.; Zheng, G. J. Anal. Appl. Pyrolysis 1991, 21, 133-146.

(21)

Di Blasi, C.; Branca, C.; Lombardi, V.; Ciappa, P.; Di Giaomo, C., Energy Fuels 2013,

27, 6781-6791. (22)

Corbetta, M.; Frassoladati, A.; Bennadji, H.; Smith, K.; Serapiglia, M.; Gauthier, G.;

Melkior, T.; Ranzi, E.; Fisher E.M. Energy Fuels 2014, 28, 3884-3898. (23)

Peters, B.; Bruch, C. J. Anal. Appl. Pyrolysis 2003, 70, 233-250.

(24)

Bennadji, H; Smith, K.; Shabangu, S.; Fisher, E.M. Energy Fuels 2013, 27, 1453-1459.

(25)

Suleiman, B.M.; Larfeldt, J.; Leckner, B.; Gustavsson, M. Wood Sci. Technol. 1999, 33,

465-473. (26)

White, C. M.; Strazisar, B. R.; Granite, E. J.; Hoffman, J. S.; Pennline, H. W. J. Air

Waste Manage. Assoc. 2003, 53, 645-715. (27)

ECN, 2015, “Phyllis2, database for biomass and waste,” https://www.ecn.nl/phyllis2,

Energy research Centre of the Netherlands, accessed 7/26/15. (28)

Smith, D.N.R., and Cockcroft, R., Nature 1961, 4759, 163-4.

(29)

Di Blasi, C.; Galgano, A.; Branca, C. Energy Fuels 2009, 23, 1045-1054.

(30)

Liu, Q.; Wang, S.; Luo, Z.; Cen, K. J. Chem. Eng. Jpn. 2008, 41, 1133-1142.

(31)

Yang, H.; Yan, R.; Chen, H.; Zheng, C.; Lee, D. H; Liang, D. T. Energy Fuels 2006, 20,

388-393. (32)

Yang, H. P.; Chen, H. P.; Du, S. L.; Chen, Y. Q.; Wang, X. H.; Zhang, S. H. Proc. CSEE

2009, 29, 70-75. (in Chinese) (33)

Park, W.C.; Atreya, A.; Baum, H. Park, W.C.; Atreya, A.; Baum, H.R. Combust. Flame

2010, 157, 481-494.

ACS Paragon Plus Environment

Page 21 of 34

Energy & Fuels

21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(34)

Gauthier, G.; Melkior, T.; Grateau, M.; Thiery, S.; Salvador, S. J. Appl. Anal. Pyrolysis

2013, 104, 521-530. (35)

Anca-Couce, A.; Mehrabian, R.; Scharler, R.; Obernberger, I. Energy Convers. Manage.

2014, 87, 687-696. (36)

Bennadji, H; Smith, K.; Serapiglia, M.; Fisher, E.M. Energy Fuels 2014, 28, 7527-7537.

ACS Paragon Plus Environment

Energy & Fuels

Page 22 of 34

22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table 1. Mass loading of potassium in wood samples, dry basis, determined with different methods. Doping Liquid

ppm K, based ppm K, based on

ppm K, based on

ppm K, based on

on wet vs dry

change in dry

ICP measurements,

ICP

cylinder mass

cylinder mass

average of spatially

measurements of

during doping

before and after

resolved

samples used for

doping

measurements

TGA.

Distilled H2O

N/A

N/A

180

129

0.900% aqueous

6300±280

5780±350 if

4640

4700

6940

7600

7220

8380

solution KCl

added mass is

(low KCl)

KCl

1.560% aqueous

11540±160

9390±260 if

solution KCl

added mass is

(high KCl)

KCl

1.444% aqueous solution K2CO3

10770±300

9230±470 if added mass is K2O

± values represent standard deviations over the 5-7 cylinders used in the experiments.

ACS Paragon Plus Environment

Page 23 of 34

Energy & Fuels

23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table 2. Yields of all measured species and char, mg/kg, dry, additive-free basis. Gas and tar yields were obtained through a trapezoid rule integral of the FTIR species data, combining multiple runs. For the char, the standard deviation of repeated measurements is reported. Char yield was not measured for the case marked “n.m.” because parts of the char broke off in the reactor and were not retrieved. The measured mass of potassium additive is subtracted from the measured solid mass to obtain the char mass for use in this table. Distilled water

KCl, low

KCl, high

K2CO3

Gases: Carbon dioxide

102.6

108.2

110.5

116.1

36.3

45.1

46.8

59.6

3.3

2.7

2.8

3.4

142.2

156.0

160.1

179.1

Acetic acid

47.0

51.5

54.9

52.4

Formaldehyde

24.6

10.5

11.7

5.8

Methanol

14.7

21.1

22.6

26.0

3.2

1.8

1.6

0.5

249.5±4.3

290.5±1.7

288.9±2.7

n.m.

Carbon monoxide Methane Sum of gases Tars:

Formic acid Char

ACS Paragon Plus Environment

Energy & Fuels

Page 24 of 34

24

2.0 mass fraction K, %

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1.5 1.0 0.5 0.0 0 1 2 3 4 axial sampling location, cm from end of cylinder

Figure 1. Spatial distribution of potassium in a wood cylinder doped with K2CO3. Each measurement is an average over an equal volume of wood. Different shadings represent different radial positions. For black,ඥ2ܴ/3 ≤ ‫ ;ܴ ≤ ݎ‬for white, ඥܴ/3 ≤ ‫ ≤ ݎ‬ඥ2ܴ/3; for striped, ‫ ≤ ݎ‬ඥܴ/3, where R is the cylinder radius and r is the sampling radius.

ACS Paragon Plus Environment

Page 25 of 34

Energy & Fuels

25

140 120 100

5

-1

dX/dt·10 (s )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

80 60 40 20 0 100

150

200

250 300 o Temperature ( C)

350

400

450

Figure 2. Rate of change of mass normalized by dry mass, versus temperature, from thermogravimetric analyzer experiments for four different wood treatments: black: distilled water; blue: low KCl; green: high KCl; red: K2CO3.

ACS Paragon Plus Environment

Energy & Fuels

Page 26 of 34

26

500 400 o

Temperature ( C)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

300 200 100 0 0

100

200

300 400 Time (s)

500

600

Figure 3. Representative centerline temperature traces. Each curve is a separate experiment; different colors correspond to different wood treatments: black: distilled water; green: high KCl.

ACS Paragon Plus Environment

Page 27 of 34

Energy & Fuels

27

500

o

Temperature ( C)

400 300 (a)

200 100 0 0

100

200

300

400

500

600

Time (s)

480 470 o

Temperature ( C)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

460 450

(b)

440 430 420 410 150 200 250 300 350 400 450 500 550 600 Time (s)

Figure 4. Temperature profiles at the center (solid lines) and at R/2, L/4 (dashed line) for four different wood treatments: black: distilled water; blue: low KCl; green: high KCl; red: K2CO3.

ACS Paragon Plus Environment

Energy & Fuels

Page 28 of 34

28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Each line is the average of results from 3 or 4 experiments. Part (b) is a closeup view, with temperatures plotted only in the vicinity of the temperature peak for each treatment.

ACS Paragon Plus Environment

Page 29 of 34

Energy & Fuels

29

6

3

o

-1

dT/dt ( C·s )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0

·

-3 0

100

200

300 400 Time (s)

500

600

Figure 5. Rate of temperature increase with time, center of particle, for four different wood treatments: black: distilled water; blue: low KCl; green: high KCl; red: K2CO3. Each line is the average of results from 3 or 4 experiments. Temperatures are smoothed (moving average) and then numerically differentiated.

ACS Paragon Plus Environment

Energy & Fuels

Page 30 of 34

-1

-1

Production Rate (mg·s ·g wood)

30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0.6 0.4 0.2

Water

0.6 0.4 0.2

Low KCl

0.6 0.4 0.2

High KCl

0.6 0.4 0.2

K2CO3

0

100

200 300 Time (s)

400

500

Figure 6. Rate of production of carbon dioxide (triangles) and carbon monoxide (circles), versus time after insertion, for four different wood treatments, on a dry additive-free basis. Results from three or more experiments for each treatment are plotted together.

ACS Paragon Plus Environment

Page 31 of 34

Energy & Fuels

31

0.4

-1

-1

Production Rate (mg·s ·g wood)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Water

0.2 0.4 Low KCl

0.2 0.4

High KCl

0.2 0.4

K2CO3

0.2 0

100

200 300 Time (s)

400

500

Figure 7. Rate of production of acetic acid versus time after insertion, for four different wood treatments, on a dry additive-free basis. Results from three or more experiments for each treatment are plotted together.

ACS Paragon Plus Environment

Energy & Fuels

Page 32 of 34

32

0.2

-1

-1

Production Rate (mg·s ·g wood)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Water

0.1 0.2 Low KCl

0.1 0.2

High KCl

0.1 0.2

K2CO3

0.1 0

100

200 300 Time (s)

400

500

Figure 8. Rate of production of methanol (triangles) and formaldehyde (squares) versus time after insertion, for four different wood treatments, on a dry additive-free basis. Results from three or more experiments for each treatment are plotted together.

ACS Paragon Plus Environment

Page 33 of 34

Energy & Fuels

-1

-1

Production Rate (mg·s ·g wood)

33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0.03 0.02 0.01

Water

0.03 0.02 0.01

Low KCl

0.03 0.02 0.01

High KCl

0.03 0.02 0.01

K2CO3

0

100

200 300 Time (s)

400

500

Figure 9. Rate of production of methane (diamonds) and formic acid (circles) versus time after insertion, for four different wood treatments, on a dry additive-free basis. Results from three or more experiments for each treatment are plotted together.

ACS Paragon Plus Environment

Energy & Fuels

Page 34 of 34

34

500 400 Time (s)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

300 200 100 0 50% gas release

95% gas T maximum T maximum (center) (R/2, L/4) release

Figure 10. Characteristic times of gas release and of the occurrence of maximum temperature, for four different wood treatments: black: distilled water; blue: low KCl; green: high KCl; red: K2CO3.

ACS Paragon Plus Environment